Unparsed ANSI codes in transcript html #95

First: thank you for making and distributing a helpful tool!

I captured a transcript of a session which ended with /context, since I was interested in doing some auditing of context and token usage. The context output in the transcript html included the raw ANSI escape codes, which made that specific response rather tough to read:

 �[1mContext Usage�[22m �[38;2;136;136;136m⛁ �[38;2;153;153;153m⛁ ⛁ �[38;2;177;185;249m⛀ �[38;2;215;119;87m⛀ �[38;2;255;193;7m⛀ �[38;2;147;51;234m⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ �[39m Opus 4.6 (1M context) �[38;2;153;153;153mclaude-opus-4-6[1m]�[39m �[38;2;147;51;234m⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ �[39m �[38;2;153;153;153m228.5k/1m tokens (23%)�[39m

No other text in the transcript showed similar problems.

Ideally, I would expect the text content to be parsed from the styled terminal output, either as plain text or, perhaps better, with equivalent styling in the output html.